Resuming a failed send

[javocado]

I'm trying to zfs send a rather large (10T) snapshot and it keeps failing (due to network disruptions and the like). I'm wondering what the status of things are at this time to be able to resume a broken transfer and that is done. I will be sending to a FreeBSD 10.3 host. If this is not currently possible, what are the other options? I am aware I can dump to a file and then rsync that file and zfs receive that file, but that is a multi-step, slow process with it's own drawbacks.

Thanks!

 

[Ericloewe]

Unfortunately, there's not much you can do. Maybe you could try piping zfs send through a more resilient transport (can't really think of any specific options, though)...

 

[javocado]

Actually, I found that I can do a:

zfs send | zfs receive -s

and if it fails I can get a token with

zfs get receive_resume_token

And resume it. The question remaining: is there ANY way to get the receive_resume_token from my previously-started transfer when I didn't use -s during the receive. i.e. a zfs list does not show any token available, so can I somehow generate one?

 

[rs225]

I think the answer is no. Without -s, a failed zfs receive drops everything it had received up to the failure.